1. Project number and Title
Team: Pasha Kazatsker
Title: Robotic Video Game Learning Using Self-Detection
Number: 7


2. Should this project be considered for the Best Project award? (yes/no)
no

3. Should this project be considered for the top 3 project awards? (yes/no)
no

4. On a scale of 1 to 10, how would you rate the overall organization/clarity
of the project report? (1-10)
6

5. On a scale of 1 to 10, how would you rate the overall project idea? (1-10)


6. On a scale of 1 to 10, how would you rate the overall research
contribution of the project idea, methodology and/or results?  (1-10)
6


Then write approximately 2 pages of helpful feedback to the project's
author(s). The following questions should help you organize your
feedback:

* Overall, is the project report clear, concise, and well-organized?
Clear: 5
The paper was riddled with overly complicated sentences and grammatical errors. Spelling mistakes (many which could have been caught by a spell checker) capitalization errors and convoluted sentences which required more sentences to explain them were found throughout the paper. It often felt as if the authors were trying to convey their intelligence through complicated sentence structure and a large vocabulary. It would have been much preferred to explain things in the simplest form possible. Audience should always be considered when writing anything. In this case audience is the 585 class, TA and professor. Speaking as part of the audience I found the writing style of the author distracting from the quality work he was performing.

Concise: 7
As mentioned previously, the paper suffered from expressing itself in more complicated ways then necessary. However, it also does a great job of getting to the point on most technical details. It is appreciated when a library function is explicitly mentions and briefly described.

Organized: 8
Organization is perfectly fine. From the table of contents the report can easily be followed and jumped to any desired location. Section flow is very good. Within a section flow sometimes leaves something to be desired. Sentences often felt broken and choppy. 

* How does the project idea and methodology fit within the framework of Developmental Robotics?
The project does something that many projects have done previously, but it does so in the physical world instead of simulation. This seems to be a pretty common research task in current developmental robotics. Although nothing groundbreaking, a lot can be said for showing that theory and application can coincide and what needs to be adjusted to make many AI methods realizable.

* Describe what you like BEST about the project?
The best part of the project is the explanation of the self detection system used. The graph used to identify between self and other was an interesting approach. The way everything was modularized was also a great idea. It will allow the project to continue to evolve, possibly with multiple contributors, in parallel.

* Describe what you like LEAST about the project?
It didn’t seem very novel. Also the project had constraints that were put in place simply because there was a problem. If something didn’t work it seemed that instead of trying to overcome an obstacle, the obstacle was stated and made a restriction. It would have been much more meaningful to try to work through the obstacles. 
For example the problem of objects disappearing sounds like a Nyquist sampling issue. The author may have tried to overcome this but if he did it was failed to mention. Reducing the frequency of the display to half the frequency of the webcam may have solved many video issues. If this wasn’t possible a filter could have been introduced on the frames to try to reduce the high frequency ‘noise’ of the objects disappearing.

* Do the methods, results and contributions of the final project correspond to what was presented in the initial project proposal?
Yes. The final project follows the proposal well.


* Are there any major details left out with regards to the methods, algorithms, or experimental design described in the report?
Not really. The design of the experiment was very well explained. From a technical standpoint the project paper was very sound. I appreciated the detail of algorithm explanation.

* Do the experimental results reported in the paper demonstrate success?
Yes.

* Do you have any suggestions for improvement and future work?
As mentioned in the paper, using a mutual information/entropy method could produce good results. However, if the object disappearing problem isn’t solved the entropy method doesn’t discriminate self from other quickly enough. So the first step is to fix the visual problems of the model.
I think the most important future work would be fix the visual artifacts.

* How close is the final project report to being publishable as a conference or journal paper (consider the research papers that were part of the course reading)? What would it take to get there?
It definitely needs fine-tuned and proof-read but the project seems close to publishable level. The report still needs edited and reviewed but those are small things compared to the research.